Abstract In macaque monkey, frontal and parasagittalbrain sections were stained with SMI-32, an antibody di-rected against a nonphosphorylated neurofilament pro-tein that labels pyramidal cells. The goal of this investi-gation was to find reliable criteria with which to drawthe border between the motor (M1) and premotor (PM)cortex and delineate subdivisions within the lateral PM.Two-dimensional reconstruction of the staining patternswas also performed by flattening the series of frontalsections. The distribution of SMI-32 immunoreactivityin layers III and V of the cortex revealed the existence ofthree subregions in the ventral rostral PM and a clearmediolateral boundary within the dorsal PM defined byclusters of SMI-32-positive pyramidal cells in layer V.The border between M1 and PM was easily distin-guished at the level of the dorsal PM by a strong loss ofimmunoreactive pyramidal cells in layers III and V. Atthe level of the ventral PM there was no clear disruptionof layer V pattern, and the border was set using the pat-tern of layer III immunoreactivity.

Key words Lateral premotor cortex · SMI-32 · Pyrami-dal cell · Motor cortex · Primate

Introduction

The premotor cortex (PM) is functionally and architec-tonically heterogeneous and has been described in vari-ous ways (Brodmann 1909; Vogt and Vogt 1919; Boninand Bailey 1947; Barbas and Pandya 1987). Patterns ofcytochrome oxidase staining and cytoarchitectonics en-abled Matelli et al. (1985, 1991) to subdivide the PM ofthe macaque monkey into six subareas. In the lateral PM

they distinguished F4 (PM ventral caudal, PMvc), F5(PM ventral rostral, PMvr), F2 (PM dorsal caudal,PMdc) and F7 (PM dorsal rostral, PMdr). Campbell andMorrison (1989), using a new antibody to nonphosphor-ylated neurofilament protein (SMI-32), demonstratedthat a subpopulation of pyramidal cells was stained inthe primate brain cortex. Recently, a series of investiga-tions demonstrated that this immunoreactivity had spe-cific regional distribution patterns. In the monkey thistechnique helped subdivide various brain regions, suchas the superior temporal cortex, visual areas, orbital andmedial prefrontal cortex and inferior pulvinar (Cusick etal. 1995; Chaudhuri et al. 1996; Hof and Morrison 1995;Carmichael and Price 1994; Gutierrez et al. 1995). Ap-plying this technique to the monkey motor cortex (M1),Preuss et al. (1997) proposed three rostrocaudal subdivisions. SMI-32 was also processed in the humancortex for delineating regions within the orbitofrontalcortex (Hof et al. 1995) and the superior and mesial PM(Baleydier et al. 1997). The latter authors were able toeasily define a caudal and a rostral subdivision in lateralarea 6. They also stated that the SMI-32 staining wasmore reliable than cytochrome oxidase because of theextreme variations in this enzymic method.

In the present study we attempted to reinvestigate thecontroversial parcellation of the lateral PM with this newtool, which should be especially appropriate for demon-strating the various distribution of layer III and V pyra-midal cells within the agranular frontal cortex.

Materials and methods

Two monkeys (Macaca mulatta and M. fascicularis) were given alethal overdose of sodium pentobarbital, perfused transcardiallywith 0.4 l normal saline followed by 4 l fixative consisting of 4%paraformaldehyde in 0.1 M phosphate buffer pH at 7.4 (PB). Per-fusion was continued with increasing concentration of sucrose upto 30% in PB. After the brain was removed from the skull, it wasstored in cold 30% sucrose in PB as a cryoprotectant, prior to sec-tioning on a freezing microtome at 50 µm. Three hemisphereswere cut in the coronal and one in the parasagittal plane. Everytenth section was processed for SMI-32 immunohistochemistry. A

L. Gabernet · M.-C. Hepp-ReymondBrain Research Institute, University Zurich,Zurich, Switzerland

L. Gabernet · V. Meskenaı..te · M.-C. Hepp-Reymond (✉)Institute of Neuroinformatics, University of Zurich-Irchel,Winterthurerstrasse 190, CH-8057 Zurich, Switzerland,e-mail: mchr@ini.phys.ethz.ch

Exp Brain Res (1999) 128:188–193 © Springer-Verlag 1999

R E S E A R C H N O T E

Laetitia Gabernet · Virginia Meskenaı..teMarie-Claude Hepp-Reymond

Parcellation of the lateral premotor cortex of the macaque monkeybased on staining with the neurofilament antibody SMI-32

Received: 30 October 1998 / Accepted: 4 March 1999

M1 could be clearly characterized by its high number ofSMI-32-positive pyramidal cells in layer V. Most of theBetz cells were labeled as could be judged from a com-parison with the Nissl-stained sections. Layer III wasalso rich in immunoreactive pyramidal cells, which weregenerally smaller and less strongly stained than those ob-served in layer V. This layer was thick but not homoge-neous in labeling. Some regions in M1, mainly in therostral part, were stained less, for both layer III and layerV. In particular there was a disappearance of the immu-noreactivity close to the CS crown (Figs. 1F, 2A). Thisdecrease was observed in all four hemispheres pro-cessed.

Border between motor and premotor cortex

We distinguished the border between M1 and PM by theabsence or strong decrease in the number of SMI-32-positive cells in layer V and by the decrease in immuno-reactivity in layer III. The M1/PM border was obvious atthe level of PMd (Figs. 1F, 2C,D, 3, left). PMd containedalmost no labeled cells in layer V and only a low densityof SMI-32-positive cells in layer III. However, severalclusters of SMI-32-positive cells in layer V were foundin the superior precentral sulcus (SPcS) and more ros-trally (Fig. 1E,F). These cell clusters can be used to de-fine a mediolateral partition within PMd (Fig. 3: 3 vs 8and 2) as the cortex in the medial direction contained athinner layer III and had weaker staining than in its later-al aspect.

The border of M1 with PMv is less clear than it iswith PMd due to the presence of many immunoreactivepyramidal cells in layer V in PMvc. At this level, SMI-32-positive pyramidal cells in layer V showed no discon-tinuity from caudal to rostral and made it difficult to de-lineate an M1/PM border on the single criterion of layer-V-labeled cells. However, this border could be set takingas criteria both the strong decrease in density and size ofthe layer V pyramidal cells and the thinner and lessdense layer III (Fig. 2A,B).

Borders within PM

The distribution of layer V SMI-32-positive pyramidalcells in PMv was a useful feature with which to drawthe limit between PMv and PMd. In PMd, layer V waspoorly labeled or mostly absent (Figs. 1E, 2C) exceptfor the region close to the spur. Within PMd two sub-divisions could also be distinguished in the rostrocau-dal direction: a caudal region containing SMI-32-posi-tive cell clusters (PMdc) and a quite rostral regionshowing very poor immunoreactivity (PMdr). In Figs.2C,D and 3, these subregions are numbered 2 and 8 re-spectively.

PMvr and PMvc can also be distinguished on the basisof the SMI-32 staining, the border being defined by thestrong decrease in the density of the layer III staining (Fig.

189

series of adjacent sections were stained for Nissl bodies with cre-syl violet, for cytochrome oxidase (CO), acetylcholinesterase(AChE), parvalbumin (PV), calretinin (CR), and the α1- and α2-subunits of the GABAA receptor.

The sections were preincubated in 20% normal goat serum in0.05 M Tris-buffered saline at pH 7.4, which contained 0.5% Tri-ton-X100 (TBST) for 60 min at room temperature. They were thenincubated in SMI-32 monoclonal antibodies (1:7500, SternbergerMonoclonals Inc., MD) for 24–36 h at 4°C in TBST also contain-ing 2% bovine serum albumin, 1% normal goat serum, and 2%normal monkey serum. Following 4×20-min washes in TBST, thesections were incubated for 12 h at 4°C in biotinylated secondarygoat anti-mouse antibodies (1:200; Vector Laboratories, Burlin-game, CA) and for 3 h in the Vectastain ABC Elite reagent (1:100,Vector Laboratories) at room temperature. Antigenic sites were vi-sualized with standard 3,3’-diaminobenzidine tetrahydrochloride(DAB; Sigma, Switzerland) histochemistry. Firstly, sections werepreincubated in 0.05% DAB in TB, pH 7.6, for 20–25 min. H2O2was then added to the media to give a final concentration of0.0048% and sections were incubated for a further 3–5 min. Thereaction was stopped by several washes in TB. Sections weremounted on gelatinized slides from PB, air dried, dehydrated, andcoverslipped in Entelan (E. Merck, Switzerland). As controlssome adjacent sections were processed simultaneously as de-scribed above, except that the primary antibody was omitted or re-placed by normal mouse serum.

SMI-32-labeled sections were examined under bright-field op-tics and compared with adjacent Nissl-stained sections. Sectionswere digitized under a camera driven by a multiple-channel imagingdevice system. Information from sections was compiled into a stackof digitized sections and the (10%) shrinkage due to perfusion (2%)and immunostaining (8%) was corrected. Flattened two-dimension-al reconstructions of frontal cortex were produced for each hemi-sphere by “opening” the central (CS) and the arcuate (AS) sulcus onevery section (adapted from Dum and Strick 1991). The portion ofthe cortex from the fundus of the AS and its caudal bank rostrally tothe tip of the intraparietal sulcus caudally was reconstructed. Usingthis method the whole lateral PM and M1 cortex was flattened.SMI-32-labeled layer III was used as the reference layer since it wasstained on the whole surface of the agranular frontal cortex. It was“straightened” mediolaterally from the hemispheric midline to thelateral sulcus on each frontal section. The boundaries within the PMwere determined microscopically, marked on the printed digitizedsections and projected to the straightened layer III. Every sectionwas aligned on the midline (Fig. 3).

Results

Four hemispheres of two monkey brains were processedwith the SMI-32 antibody. SMI-32 labeled a populationof pyramidal cells in layers III and V. SMI-32-positivelarge pyramidal cells could be identified when comparedwith adjacent Nissl-stained sections. To set boundaries,several nonquantitative criteria were applied such as thedensity of layer III and discontinuities in the presenceand number of layer V SMI-32-positive pyramidal cells.The boundary of labeled SMI-32-positive large pyrami-dal cells in layer V was the main criterion for drawingthe border between M1 and PM. To set boundaries with-in PM the density of the staining in layer III assessed byvisual observation was the main criterion.

Staining in the motor cortex

The staining in M1 and PM was quite strong comparedwith other adjacent cortical areas (Fig. 1F,G). Staining inM1 appeared at the level of the pyramidal cell bodies,and both apical and basal dendrites in layers III and V.

190

2A,B). PMvr (F5) was generally weakly stained, withoutany layer V SMI-32-positive cells, except in a regionclose to the spur. Parasagittal sections allowed us to definemore clearly three PMvr subdivisions labeled 7, 6 and 5 inFig. 2A–C. Firstly, the most rostral and lateral region (7)showed almost no SMI-32 staining, except some pale pro-cesses in layer III. An adjacent caudal part (6) contained athin weak but clear SMI-32-positive layer III. The thirdsubdivision (spur region, 5) had a thick and highly immu-noreactive layer III and contained scattered large pyrami-dal cells in layer V (Fig. 1B–D and inset B’). These largeSMI-32-positive layer V neurons around the spur are alsovisible in the sagittal sections of Fig. 2.

To obtain a precise and clear overview of the stain-ing pattern within the AS, we opened the sulcus by flat-tening the series of frontal sections and reconstructing

Fig. 1A–G SMI-32 staining in a series of frontal sections throughthe monkey agranular frontal cortex (Macaca mulatta). Pyramidalcells in layers V and III are labeled. SMI-32 immunoreactivityshows distinct labeling patterns in the primary motor (M1) andpremotor (PM) areas. On the right is a sketch showing the locationof the sections from rostral (A) to caudal (G) (same hemisphere asFig. 3, left). Arrows indicate boundaries set according to the pres-ence and the density of labeled pyramidal cells in layers III and Vbetween: PMvr (PM ventral rostral), PMvc (PM ventral caudal),PMdr (PM dorsal rostral), PMdc (PM dorsal caudal) and the bor-der between PMdc and M1. Arrowheads indicate the presence oflayer V SMI-32-positive cells (IAS inferior arcuate sulcus, SAS su-perior arcuate sulcus, CS central sulcus, SPcS superior precentralsulcus). Inset (B’) enlargement of the spur region in section Bshowing scattered large pyramidal cells in layer V

191

the surface of the stained cortex. The opening of the CSand AS created distortions on the reconstructed maps atthe level of the AS spur and the CS foot, which can beappreciated as strong irregularities in the course of thelateral sulcus (Fig. 3). These flattened reconstructionsrevealed the subdivisions within PMvr, and the variouspossible boundaries in PMd. They also allowed a com-parison between animals and hemispheres. These mapsclearly show the continuous presence of layer V pyra-midal cells from the CS to the AS at the level of thespur. They are still preliminary as they do not includethe rostrocaudal gradient in density and size of theSMI-32-stained pyramidal cells, which was used todraw the line between M1 and PMvc. In the drawingsrepresenting the lateral surface of the two flattenedbrains, an attempt has been made to trace the bordersand show the parcellation of PM discussed in the text.Unfortunately this representation is incomplete on thissurface view of the brain, as some borders are located

within the sulci and therefore cannot be represented ona surface map.

Discussion

The distribution of neurofilament protein with SMI-32antibody displayed different patterns of immunoreactivi-ty within the agranular frontal cortex. On the basis of theimmunoreactivity in layer III and V pyramidal cells, wewere able to set a clear border between M1 and PM,even at the level of PMvc, which still contains SMI-32-positive pyramidal cells. Boundaries in this very regionhave often not been easy to define, although Matelli etal. (1985) found some criteria with which to set theirF1/F4 border on the basis of CO staining. Our M1/PMborders, which are supported by adjacent sectionsstained for CO, AchE and Nissl (Gabernet et al. 1997),seem to correspond with the ones proposed by Matelli etal. (1985). In addition, striking differences between theimmunoreactive patterns of PMd as compared to PMvwere obvious mainly on the frontal sections. In PMd wealso found a clear mediolateral boundary defined by theclusters of SMI-32-positive pyramidal cells in layer V atthe level of the SPcS and more rostrally. Although thepoor immunoreactivity in the most dorsal PMd could beattributed in Fig. 1 to inhomogeneity in the section thick-ness, we do not think that it is the case as similar gradi-ents have been clearly observed in two other hemi-

Fig. 2A–D SMI-32 staining in a selection of parasagittal sectionsthrough the monkey frontal cortex (Macaca mulatta). Centersketch of the left hemisphere showing the location of the sectionsfrom ventral (A) to dorsal (D). Arrows indicate boundaries withinthe lateral PM and between PM and M1 according to the criteriadescribed in Fig. 1. Numbers correspond to the parcellation of thelateral PM shown in the sketches in Fig. 3 and discussed in thetext (2 PMdc, 4 PMvc, 5 PMvr spur, 6 PMvr intermediate, 7 PMvrrostral, 8 PMdr). Stars indicate layer V pyramidal cells in PMdr,close to the AS internal spur. For abbrevations see Fig. 1

192

spheres (Fig. 3). A mediolateral partition of PMd hasalso been suggested on the basis of SMI-32 immunoreac-tivity by Geyer et al. (1998).

Rostrocaudal subdivisions within PMd (8, 2) could bebest seen on parasagittal sections and may correspond tothe F2 and F7 of Matelli et al. (1991). The caudal regionwith cell clusters in the SPcS was defined as belongingto PMdc. Preuss et al. (1997), because of the presence oflayer V pyramidal cells, decided that this region still be-longed to M1 as a third rostral region (4r). As can beclearly seen in Fig. 3, discontinuities in layer V SMI-32

Fig. 3 Top Flattened reconstructions of SMI-32-staining patternsin two monkeys (left M. mulatta, same as Fig. 1; right M. fascicul-aris). Flattened sections aligned on the midline (junction of themedial wall with the lateral surface of the cortex) [dashed linesfundus of each opened sulcus, full lines crowns of sulci, except forthe most caudal line of AS which marks the end of the internalspur (section displayed in Fig. 1D), dark gray region of the cortexwith dense-labeled layer III and layer V pyramidal cells, middlegray cortical regions with labeled layer III only, pale gray portionsof PM with weak staining in layer III]. Bottom Lateral view of thecorresponding brains with a drawing of the subdivisions discussedin the text (1 M1, 2 PMdc, 3 PMd medial, 4 PMvc, 5 PMvr spur,6 PMvr intermediate, 7 PMvr rostral, 8 Pmdr)

193

labeling between the CS and the SPcS strongly suggestthat this region is not an extension of M1 but rather be-longs to PMd. In fact, we showed in two monkeys thatlow-current intracortical microstimulation (ICMS) elicit-ed finger movements in the region of the clusters, reveal-ing a new PM finger region localized more rostrally andseparate from the caudal M1 hand representation (Qi etal. 1994). Our localization of the M1/PMd border agreeswith the generally accepted one (He et al. 1993).

In PMv, although the boundary between rostral andcaudal regions (F4/F5) was not sharp everywhere, subdi-visions could be distinguished. PMvc was defined ashaving a staining pattern rather similar to that of M1, butwith less numerous cells in layer V and a fainter layerIII. This part of the PMv could not be additionally subdi-vided mediolaterally as reported by Geyer et al. (1998).An interesting finding is that the PMvr (F5) could besubdivided into three parts, as previously suggested byMatelli et al. (1996) on the basis of various stainings.The most rostral and lateral one was without any SMI-32immunoreactivity but with a semigranular layer IV andparticularly well laminated staining patterns in adjacentsections stained for AChE, PV, CR, CO and α1- and α2-subunits of the GABAA receptor (Fig. 3: 7). A caudal ad-jacent one had a weak SMI-32 immunoreactivity in layerIII, and the most caudal and medial region at the level ofthe spur contained striking immunoreactivity in layer V(Fig. 3: 6 and 5 respectively). This small region can becompared with the region found in the fundus of theSPcS and looks like an island of scattered immunoreac-tive layer V pyramidal cells, similar but independent ofM1 proper. This point is corroborated by functional datashowing that within the arcuate spur, and also lateral toit, ICMS could elicit finger movements, even at relative-ly low currents (Hepp-Reymond et al. 1994; Gabernet etal. 1997). For one of the hemispheres processed in thepresent study, the ICMS sites have also been reconstruct-ed on a flattened map that could be compared with theanatomical one. We found that the PM sites with thelowest current threshold for eliciting finger movementscould in part be superimposed on the islands of SMI-32-positive pyramidal cells (unpublished data). These is-lands could also correspond to clusters of corticospinalneurons described by Dum and Strick (1991) and He etal. (1993).

In conclusion, the distribution of SMI-32 is a usefultool with which to parcellate the M1 and PM areas. Al-though the lateral PM cortex can be quite variable frommonkey to monkey at the macroscopic and microscopiclevels, some general features could be assessed to definevarious anatomical subdivisions within this region.

Acknowledgements This study was supported by the Swiss National Research Foundation, grant no. 31-39679.93, the National Research Program 38, grant no. 40-52837/1, and theSchweizerische Hochschulrektorenkonferenz. We thank Dr. P. Streit for help with the MCID system, Dr. A. McKinney for criti-cally reading the manuscript and D. Latawiec for expert technicalassistance.

References

Baleydier C, Achache P, Froment JC (1997) Neurofilament archi-tecture of superior and mesial premotor cortex in the humanbrain. Neuroreport 8:1691–1696

Barbas H, Pandya DN (1987) Architecture and frontal corticalconnections of the premotor cortex (area 6) in the rhesus mon-key. J Comp Neurol 256:211–228

Bonin G von, Bailey P (1947) The neocortex of Macaca mulatta.University of Illinois Press, Urbana, IL

Brodmann K (1909) Vergleichende Localizationslehre der Gross-hirnrhinde in ihren Prinzipien dargestellt auf Grund desZellenbaues. Barth, Leipzig

Campbell MJ, Morrison JH (1989) Monoclonal antibody to neuro-filament protein (SMI-32) labels a subpopulation of pyramidalneurons in the human and monkey neocortex. J Comp Neurol282:191–205

Carmichael ST, Price JL (1994) Architectonic subdivision of theorbital and medial prefrontal cortex in the macaque monkey. JComp Neurol 346:366–402

Chaudhuri A, Zangenehpour S, Matsubara JA, Cynader MS(1996) Differential expression of neurofilament protein in thevisual system of the vervet monkey. Brain Res 709:17–26

Cusick CG, Seltzer B, Cola M, Griggs E (1995) Chemoarchitec-tonics and corticocortical terminations within the superiortemporal sulcus of the rhesus monkey: evidence for subdivi-sions of superior temporal polysensory cortex. J Comp Neurol360:513–535

Dum RP, Strick PL (1991) The origin of corticospinal projectionsfrom the premotor areas in the frontal lobe. J Neurosci11:667–689

Gabernet L, Qi H-X, Arnold M, Hepp-Reymond M-C (1997) Sub-divisions in the lateral premotor cortex revealed by various an-atomical criteria and functional correlates. Soc Neurosci Abstr23:1274

Geyer S, Matelli M, Luppino G, Zilles K (1998) A new micro-structural map of the macaque monkey lateral premotor cortexbased on neurofilament protein distribution. EJN 10 Suppl10:83

Gutierrez C, Yaun A, Cusick CG (1995) Neurochemical subdivi-sions of the inferior pulvinar in macaque monkeys. J CompNeurol 363:545–562

He SQ, Dum RP, Strick PL (1993) Topographic organization ofcorticospinal projections from the frontal lobe: motor areas onthe lateral surface of the hemisphere. J Neurosci 13:952–980

Hepp-Reymond M-C, Hüsler EJ, Maier MA, Qi H-X (1994)Force-related neuronal activity in two regions of the primateventral premotor cortex. Can J Physiol Pharmacol 72:571–579

Hof PR, Morrison JH (1995) Neurofilament protein defines re-gional patterns of cortical organization in the macaque mon-key visual system: a quantitative immunohistochemical analy-sis. J Comp Neurol 352:161–186

Hof PR, Mufson EJ, Morrison JH (1995) Human orbitofrontal cor-tex: cytoarchitecture and quantitative immunohistochemicalparcellation. J Comp Neurol 359:48–68

Matelli M, Luppino G, Rizzolati G (1985) Patterns of cytochromeoxidase activity in the frontal agranular cortex of the macaquemonkey. Beh Brain Res 18:125–136

Matelli M, Luppino G, Rizzolati G (1991) Architecture of superiorand mesial area 6 and the adjacent cingulate cortex in the ma-caque monkey. J Comp Neurol 311:445–462

Matelli M, Luppino G, Govoni P, Geyer S (1996) Anatomical andfunctional subdivisions of inferior area 6 in macaque monkey.Soc Neurosci Abstr 22:2024

Preuss TM, Stepniewska I, Jain N, Kaas JH (1997) Multiple divi-sions of macaque precentral motor cortex identified with neu-rofilament antibody SMI-32. Brain Res 767:148–153

Qi H-X, Hüsler EJ, Alig I, Hepp-Reymond M-C (1994) Simpleand complex relations of cortical activity in the alert monkey.Soc Neurosci Abstr 20:60

Vogt O, Vogt C (1919) Allgemeine Ergebnisse unserer Hirnfor-schung. J Psychol Neurol (Leipzig) 25:277–462